Psychosocial Etiology Research

1
Psychosocial Etiology Research

• Does a psychosocial variable X affect a disorder
  Y?
• Risk
• Age of onset, presentation, course
• How much difference does X make?
• Whats the mechanism?

2
Common and Suitable Designs

• Case-control comparison
• Retrospective (very common)
• Prospective (rarely used)
• Longitudinal-developmental prospective
  study---increasingly common
• Behavior-genetic study designs (e.g., adoption
  essentially never used)

3
Problems with Retrospective Studies

• Sampling problems
• How do they find cases controls?
• Whos willing/likely to participate?
• Measurement problems
• How do they measure factor X? (Recall)
• Causal inference problems
• Nature-nurture confounds
• Other third-variable confounds (environmental)

4
Measurement Problems

• Quasi-random recall errors
• Substantial error rate in recalling even major
  events in past year (e.g., medical hospital
  stays)
• Recall biases
• Over-recalling (GPA, possible false memories of
  sexual abuse)
• Under-recalling (income)

5
Measurement Problems, contd

• Reporting biases
• Under-reporting sensitive topics
• Potential over-reporting
• borderline patients, paranoid patients
• Question wording problems
• Easily misunderstood terms (e.g., sexual abuse)
• Confusing questions
• Was there ever a time when you couldnt remember
  X?

6
Causal Inference Problems

• Nature-nurture confounds
• Schizophrenogenic mother, or schizophrenia-predisp
  osing genes?
• Criminality in the home
• direct cause of childs criminal behavior? vs.
• criminal parents pass on criminal genes?

7
Causal Inference Problems, contd

• One environmental factor as proxy for another.
  Examples
• Poverty may be proxy for a broad range of
  associated factors (e.g., chaotic home,
  malnutrition)
• Abuse may be proxy for neglect, or vice versa

8
Common Research Topics

• Childhood adversity variables
• Chronic poverty, single parent, bad
  neighborhood, neglect, physical illness/handicap
• Episodic abuse (physical, sexual), death of
  parent, divorce of parents
• As causes of
• Childhood maladjustment (acting out, poor school
  performance, peer relations)
• Psychopathology (e.g., ADHD, CD, depression)

9
Longitudinal-Developmental Studies

• Find cohort at high risk
• Impoverished kids, abused/neglected kids (from
  CPS)
• Chronically ill kids
• Suitable control group (whats suitable?)
• Follow over time (to old age?)
• (Relatively) frequent follow-ups
• Comprehensive assessments of domains
• Challenge keep track of families, keep in study

10
Poverty and Mental Illness

• Older studies indicated association
  (schizophrenia)
• Causal inference problem downward drift vs.
  direct causation
• Best evidence (Kohn) 2/3 downward drift
• Need prospective study to sort out causation

11
ECA Poverty Substudy

• Overall design of ECA study
• Design of follow-up substudy
• One-year blind-reinterview with DIS
• Select those who had no diagnosis at Time 1
• Compare those in poverty vs. not at Time 1, on
  diagnoses at Time 2
• Huge N

12
ECA Poverty Substudy Findings

• 2X increase in rates of essentially all
  diagnoses, associated with poverty
• Exception schizophrenia---BIG increase
• May be due to sampling error
• Very small N of schizophrenia onsets in follow-up
  interval

13
Life Events in Depression

• Older studies indicated association
• Overall stress level (weighted sum of life
  events)
• Specific life events
• Situational depression concept
• Causal ambiguity
• Could (incipient) depression be causing adverse
  life events? (Time order)
• Nature/nurture confound
• Possible reporting biases (search after meaning)

14
Camberwell Study(Brown Harris)

• Retrospective study
• Patients (clinic/hospital) vs. controls women in
  poor/working class area of London
• Interviewed about difficulties, life events
• Followed back for events prior to onset, 12 mo.
• Compute brought-forward time (model of time to
  onset, assuming everyone gets depressed
  eventually)
• Careful study, but design still doesnt dispel
  causal ambiguity problems with brought-forward
  time

15
Garmezy/MastenProject Competence

• Prospective longitudinal study
• Identify cohort at risk, compare to controls
• Follow densely (e.g., every year or more often)
• Orientation stress resistance

16
Project Competence Samples

• Community Cohort Sample (N200, parents gave info
  about exposure to stressful life events)
• Infant/Early Childhood Cohort (N32 children,
  life-threatening congenital heart defect)
• Physically Handicapped Cohort (N29 severely
  handicapped, mainstreamed into regular classroom)

17
Project CompetenceAssessments

• Stress
• life events questionnaire
• interview-based method
• inference from SES
• School competence
• academic achievement
• classroom behavioral competence
• interpersonal competence
• IQ

18
Project Competence Assessments, contd

• Interviews for competence (3 factors)
• child academically oriented, psychologically
  minded, broad interests, not isolated, close
  relationship with sibs/parents, high degree of
  curiosity, good sense of humor
• child gets in trouble (school and home),
  irresponsible (chores), not conscientious,
  frequent fights/disagreements with others,
  physically aggressive
• various levels of self-esteem, levels of
  aspiration/striving/achievement motivation

19
Project Competence Advantages/Limitations

• Advantages
• Prospective, developmental perspective
• Comprehensive assessments
• Limitations
• Limited N, due to high cost per subject
• Problems keeping subjects in study
• Project Competence very successful over 90
  retention!)
• Lack of focus on severe psychopathology
• Nature/nurture confound
• Results are mostly pending